IDENTITY OF HYDROLOGIC HOMOGENOUS REGIONS USING FUZZY C-MEANS METHOD

Earth science methods for mapping, discretization, and separation of homogenous regions are recently used extensively to estimate the natural landscape response to hydrologic processes. For more homogenous study areas, the accuracy of the precision increases and the operation of water resources management system and/or natural wizard controls are less complicated. There are several methods to natural landscape discretization to identify homogenous regions. Because of continuity in landscape specifications, the use of Fuzzy classification/ clustering methods such as Fuzzy C- Mean (FCM) can be a desirable choice. In this study the FCM was used for discretization of homogenous regions in regards to runoff generation for Walnut Gulch Experimental Watershed (WGEW) located in southeast Arizona. This watershed is selected to benefit from the quality and adequacy of the hydrologic data. The process of choosing and preparation of data was addressed earlier in texts. The Principal Components Analysis (PCA) was performed and then the three essential steps for making decisions based on the FCM method was taken; determination of appropriate fuzzifier; controlling of Confusion Index (CI), and optimization of number of clusters using validation indices. It was shown that use of these validation indices would surely not lead to determination of optimal number of clusters and therefore a rational method was proposed.